Predator-Specificity of Multimodal Alarm Signals in the Eastern Gray Squirrel (Sciurus carolinensis)

Communication is a fundamental biological phenomenon, necessary in many species for successful reproduction, social interactions, and predator avoidance. Understanding animal communication systems is therefore vital to our understanding of biology. One of the core attributes of any signaling system is the information contained in each signal. In this dissertation I first review the literature on the information contained in sciurid alarm calls, and note the dearth of knowledge about the presence or absence of predator-specific alarm calls in arboreal squirrels. Arboreal squirrels are closely related to ground squirrels, which have a variety of alarm calling systems, but seek refuge in trees like many primates that have been shown to use predator-specific calls. I then examine the alarm-signaling system of the eastern gray squirrel. Using field experiments I show that some, but not all, of their alarm signals contain information about predator type, specifically, whether a predator is approaching from the air or approaching on the ground. Unlike most studies of alarm signaling, which consider only alarm calls, I considered two signaling modalities, vocal signals and tail signals. When examining the entire alarm-signaling bout, the use of vocal and tail signals is associated with threat type. Of the three vocal alarms (kuks, quaas, and moans), only moans show predator specificity, being highly associated with aerial threats. Of the two tail signals used as alarms, twitches and flicks, only flicks show predator specificity, being associated with terrestrial threats. This is a unique case of two modalities being used to specify two threat types. When gray squirrels use alarm vocalizations, their alarm calling bouts can consist of kuks, quaas, and moans. Each calling bout could consist of one, two or all three vocal alarms. To examine the information contained in alarm calls about whether a threat is approaching aerially or terrestrially, I tested alarm calling bouts for an association of each signal type with threat type (aerial or terrestrial). If alarm calls function to communicate with conspecifics, the initial period of calling should be most relevant to squirrels seeking safety, so I focused on the initial 60s of calling. In this initial period the presence of kuks is associated with terrestrial threats, as is the presence of quaas. Moans are exclusively used in response to aerial threats. Initial rates of kuks, moans, and calls in general are also associated with threat type, but rate of quaas is not. Kuks and quaas are usually mixed within calling bouts and are both associated with terrestrial threats. I then examined squirrels’ responses to playbacks of modified calling bouts to test whether kuks and quaas elicit different degrees of antipredator behavior. Kuks and quaas appear to have a similar effect on conspecific behavior, although rate of calling may have a strong impact on response. Additionally, white noise bursts of equal duration to alarm vocalizations appear to be functionally equivalent to kuks or quaas; any sudden, broadband noise may be sufficient to increase alert behavior. This broad acoustic criterion for eliciting a response may facilitate eavesdropping on heterospecific alarm calls, which are often abrupt, broadband sounds. In combination, this work demonstrates an alarm-signaling system in which signalers use varimodal (either unimodal or multimodal) signals, with the amount of information about predator type contained in their signal varying from general alarms to highly specific alarms. This study is the first to test for predator specificity in multiple signaling modalities; the differential use of tail signals as terrestrial threat alarms and vocal signals as aerial threat alarms highlights the importance of examining multiple signaling modalities in other species